Previously, the tokenizer tags were saved into the data. The support for LLaMA-Factory was very poor. Has it been fixed?
· Sign up or log in to comment
